Apparatus and method for compressing measurement data correlative to machine status

• US 6,026,348 A
• Filed: 10/14/1997
• Issued: 02/15/2000
• Est. Priority Date: 10/14/1997
• Status: Expired due to Term

• Current Assignee: BN Corporation LLC
• Original Assignee: Bently Nevada Incorporated (General Electric Company)
• Inventors: Hala, Roger A.
• Primary Examiner(s): WACHSMAN, HAL D
• Application Number:
  US08/949,905
• Time in Patent Office: 854 Days
• Field of Search: 702/56, 702/32-34, 702/38, 702/41-44, 702/54, 702/70-72, 702/76, 702/77, 702/79, 702/103-106, 702/124, 702/113-115, 702/126, 702/141, 702/180, 702/183-187 , 702/189-191, 702/193-195, 702/198, 702/FOR 123-FOR 126, 702/FOR 135, 702/FOR 136, 702/FOR 168, 702/FOR 171, 702/35, 364/474.16, 364/474.17, 364/474.19, 364/728.03, 364/178, 364/179, 364/528.14, 364/528.15, 324/76.13, 324/76.15, 324/76.22, 324/76.33, 340/683, 340/870.16, 340/679, 340/680, 73/583, 73/579, 73/116, 73/659, 73/660, 73/570, 73/577, 73/578, 73/66, 73/582, 73/618, 73/457, 73/462, 73/1.14, 73/778, 73/DIG. 1, 704/205, 704/500, 704/501, 704/503, 704/504, 704/203, 704/224, 704/229
• US Class Current: 702/56
• CPC Class Codes: G01H 1/003 of rotating machines G01H1/10 takes precedence
• CPC Class Codes: G01N 29/4427 with stored values, e.g. threshold values
• CPC Class Codes: G01N 29/46 by spectral analysis, e.g. Fourier analysis or wavelet analysis spectral signal processing per se G06F17/14

Abstract

051545044

A method and apparatus for compressing, storing and transmitting measurement data correlative to machine status is disclosed in which the measurement data is continuously sensed, sampled and processed to extract significant spectral elements including magnitude and phase information from each successive period of the originally measured data and to store those spectral elements in a memory means from an initial period of significant spectral elements and each successive period of significant spectral elements which have changed since the previous period for developing a compressed data history correlative to a continuous history of the status of the machine being monitored and from which continuous signals can be regenerated and analyzed for any earlier historical time.

Claims

1. A signal processing method for processing machinery signals correlative to machine status, the steps including:

sensing signals correlative to machine status;

converting said signals into digital values;

storing a series of said digital values into a packet defining a previously stored packet;

comparing a subsequent packet of said digital values with said previously stored packet of said digital values;

storing only said digital values included in said subsequent packet which are anomalous because they differ by a criteria from comparable digital values of said previously stored packet of said digital values;

transmitting a signal correlative to said previously stored packet of digital values and to said subsequent packet of said digital values including flagging anomalous data in said subsequent packet of digital values.

2. The method of claim 1 further including the step of replacing corresponding digital values in said previously stored packet with the anomalous data from said subsequent packet for creating a new packet used for comparing further subsequent packets.

3. The method of claim 2 further including the step of indexing each packet of said digital values as they are stored and storing an indexing signature correlative with each said packet.

4. The method of claim 3 further including the step of transmitting said indexing signature when transmitting said signal correlative to said previously stored packet of digital values.

5. The method of claim 4 wherein said indexing signature includes information concerning the location of the digital values within each packet.

6. The method of claim 5 wherein said indexing signature includes the time of occurrence of the anomalous data.

7. The method of claim 6 wherein the step of sensing signals correlative to machine status includes sensing the signals at periodic intervals.

8. The method of claim 7 wherein the step of converting said signals into digital values further includes the steps of sampling said signals with a sampling means and transforming said signals into spectral elements via a fast fourier analysis algorithm.

9. A machine vibration signal processing method, the steps including:

sampling a vibration signal;

subjecting said vibration signal to a transformation means for transforming said vibration signal into a series of spectral elements;

comparing said series of spectral elements against a criteria for retention and storing in a memory means said spectral elements which pass said criteria defining a previously stored series of spectral elements;

sampling a subsequent vibration signal;

subjecting said subsequent vibration signal to said transformation means for transforming said subsequent vibration signal into a subsequent series of spectral elements;

comparing said subsequent series of spectral elements to said criteria for retention and then comparing said subsequent series of spectral elements which have passed said criteria for retention with said previously stored series of spectral elements;

storing in said memory means spectral elements of said subsequent series which have passed said criteria for retention and which differ from said spectral elements of said previously stored series by a pre-determined amount.

10. The method of claim 9 including the step of reconstituting a continuous vibration signal from said stored spectral elements of said subsequent vibration signal and from all remaining spectral elements of said previous series which do not differ from said subsequent series by said pre-determined amount.

11. The method of claim 9 wherein the steps of sampling the vibration signal and sampling the subsequent vibration signal are both at a periodic interval.

12. The method of claim 11 further including the step of retaining both a magnitude value and a phase value for each spectral element stored in said memory means.

13. The method of claim 12 further including the step of indexing spectral elements which have passed said criteria for retention with a real-time of occurrence.

14. The method of claim 9 further including the step of re-comparing said spectral elements stored in said memory means with said criteria for retention and removing from said memory means said spectral elements which fail said criteria for retention.

15. An apparatus for compressing data correlative to continuous machine vibrations signals, comprising, in combination:

at least one sensor operatively coupled to a machine for sensing machine vibration in the form of continuous electrical signals;

sampling means adapted to receive from said at least one sensor continuous electrical signals and including means for converting into digital values said continuous electrical signals;

a control circuit commanding said sampling means to convert said electrical signals into digital values;

means for uniquely tagging each digital value with a real time value for identifying a time in history of an occurrence of when each instantaneous value of each continuous electrical signal correlative to each digital value was sensed by said at least one sensor;

means for storing in a memory means a periodicity of said digital values;

means for comparing at least one of said periodicity of said digital values with a subsequent periodicity of said digital values to determine anomalous digital values, and

means for compressing data by storing in said memory means only said anomalous digital values of said subsequent periodicity of said digital values which have changed from at least one of said periodicity of said digital values.

16. A machine vibration signal processing method, the steps including:

sampling continuous vibration signals from a machine;

transforming said vibration signals into discrete digital values;

storing at least one packet of digital values in a memory means and storing a marker in said memory means for identifying the position and the time of capture of each digital value;

comparing a subsequent packet of digital values with at least one of said packet of digital values;

storing digital values of said subsequent packet of digital values which differ from at least one of said packet of digital values and storing markers for identifying the position and time of capture of each differing value;

comparing further subsequent packets of digital values with previously stored packets of digital values;

storing digital values of said further subsequent packets of digital values which differ from said previously stored values and storing markers in said memory means identifying the position and time of capture of each differing value;

reconstituting any stored packet of digital values into a continuous wave form from the values stored for said packet being reconstituted and from the values of at least one previously stored packet which correspond to values missing from said packet being reconstituted.

17. A signal compression method for storing historical data correlative to devolving machine status, the steps including:

sensing signals correlative to machine status;

sampling said signals;

transforming said sampled signals into a series of spectral elements defining a first reading;

determining dominant spectral components of a periodicity of said series of spectral elements based on a pre-determined criteria;

storing said periodicity of dominant spectral components defining a previously stored periodicity;

determining dominant spectral components of a subsequent periodicity of said series of spectral elements based on said pre-determined criteria;

comparing said subsequent periodicity of dominant spectral components with said previously stored periodicity of dominant spectral components to ascertain devolvement in respect of spectral components correlative to machine status;

compressing said subsequent periodicity of dominant spectral components by transmitting to and storing in a host computer said spectral components of said subsequent periodicity which change from said spectral components of said previously stored spectral components and when the changes occur;

defaulting to resensing in the absence of devolvement;

generating a continuous signal correlative to machine status by transforming said stored components of said subsequent periodicity and said stored components of said previously stored periodicity which fails to differ from said subsequent periodicity of dominant spectral components.

18. A method for compressing data from waveforms characterizing machine vibration, the steps including:

successively producing a series of samples representing generally instantaneous values of a series of waveforms at time-spaced intervals;

transforming a first period of said samples of a first waveform included in said series of waveforms into a first set of spectral elements;

storing only those elements in said first set of spectral elements which are significant based on a predetermined criteria;

transforming said series of samples of each subsequent waveform included in said series of waveforms into a series of subsequent sets of spectral elements;

compressing data by successively storing only those elements in a first subsequent set of spectral elements included in said series of subsequent sets of spectral elements which have changed since the last stored elements in said first set of spectral elements, and

compressing data by successively storing only those elements in each said subsequent set of spectral elements following said first subsequent set which have changed since the last stored elements in said series of subsequent sets of spectral elements.

19. A method for compressing measurement data correlative to machine status, the steps including:

sensing machine data correlative to machine status;

sampling said data;

transforming said sampled data into spectral elements;

comparing said spectral elements to a user definable criteria for retention;

storing a series of said spectral elements which have passed said criteria for retention into a current interval of spectral elements;

comparing a subsequent series of said spectral elements which have passed said criteria for retention with said current interval of spectral elements;

compressing measurement data correlative to machine status by storing in a memory means those spectral elements from said subsequent series which have passed said user definable criteria for retention and which are anomalous because they differ by a user definable amount from comparable elements of said current interval of spectral elements.

20. The method of claim 19 wherein the step of comparing said spectral elements to a user definable criteria for retention includes the step of selecting a fixed quantity of largest spectral elements over a user definable period of spectral elements for the user definable criteria for retention.

21. The method of claim 19 wherein the step of comparing said spectral elements to the user definable criteria for retention includes selecting a percentage of the largest elements in a user definable period as the user definable criteria for retention.

22. The method of claim 19 wherein the step of sampling said data includes the step of sampling said data into discrete digital values synchronously with machine speed.

23. The method of claim 19 wherein the step of sampling said data includes the step of sampling said data into discrete values asynchronously with machine speed.

24. The method of claim 19 further including the step of sensing a mechanical phase angle of a rotating shaft of a machine in which data is being sensed and further including the step of creating a mechanical phase reference demarcation used to relate the mechanical phase angle of the rotating shaft to the sensed machine data correlative to machine status.

25. The method of claim 19 further including the step of sensing a mechanical phase reference mark and further including the step of demarcating the sensed machine data into periods having a length correlative to the sensed mechanical phase reference mark.

26. A method for compressing measurement data correlative to machine status, the steps including:

sensing machine data correlative to machine status;

sampling said data;

transforming said sampled data into spectral elements;

comparing said spectral elements to a user definable criteria for retention;

compressing measurement data correlative to machine status by storing in a memory means those spectral elements which have passed said user definable criteria for retention wherein said stored elements are correlative to machine status;

wherein the step of comparing said spectral elements with said user definable criteria for retention includes the steps of selecting a minimum element amplitude which is slightly larger than noise spectral amplitudes imbedded in the sensed machine data and retaining only those elements which have an amplitude greater than the user definable minimum element amplitude for reducing data volume and eliminating noise.

27. A method for compressing measurement data correlative to machine status, the steps including:

sensing machine data correlative to machine status;

sampling said data;

transforming said sampled data into spectral elements;

comparing said spectral elements to a user definable criteria for retention;

compressing measurement data correlative to machine status by storing in a memory means those spectral elements which have passed said user definable criteria for retention wherein said stored spectral elements are correlative to machine status;

further including the step of sensing a mechanical phase angle of a rotating shaft of a machine in which data is being sensed and further including the step of creating a mechanical phase reference demarcation used to relate the mechanical phase angle of the rotating shaft to the sensed machine data correlative to machine status;

wherein the step of sensing the mechanical phase angle includes the step of sensing a mechanical phase reference mark once per revolution of the shaft, and

further including the step of demarcating the sensed machine data into periods having a length correlative to the sensed mechanical phase reference mark.

28. A method for compressing measurement data correlative to machine status, the steps including:

continuously sensing cyclic machine vibration from at least one machine in the form of electrical signals correlative to machine status;

sensing at least one mechanical phase reference mark on a rotating shaft of at least the one machine;

relating the at least one mechanical phase reference mark of the rotating shaft to the electrical signals of machine vibration wherein a mechanical angle defines intervals of electrical signals;

sampling the machine vibration electrical signals under the orchestration of a control signal;

transforming said sampled signals into spectral elements;

converting said spectral elements into intervals of amplitude and phase elements;

comparing both the amplitude and phase of each subsequent interval of amplitude and phase elements with a previous interval of amplitude and phase elements to determine if the amplitude and/or phase of each subsequent element has changed more than a user definable amount from the element of a previous interval;

communicating the changed elements to a remote site for compressing measurement data correlative to machine status by storing only the changed elements.

29. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via a radio frequency link.

30. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via an internet link.

31. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via a microwave link.

32. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via a satellite link.

33. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via a fiber optics link.

34. The method of claim 28 wherein the step of communicating the changed elements includes the step of communicating the changed elements via a direct wire link.

35. A device for compressing measurement data correlative to machine status, said device comprising in combination:

at least one sensor operatively coupled to a machine for sensing data correlative to machine status;

sampling means operatively coupled to said at least one sensor for sampling said sensed data into discrete elements;

processor means operatively coupled to said sampling means for transforming said discrete elements into spectral elements and converting said spectral elements into magnitude and phase elements;

means for defining intervals of the magnitude and phase elements;

means for storing a current interval of the magnitude and phase elements;

means for comparing a subsequent interval of magnitude and phase elements with said current interval of magnitude and phase elements;

means for storing only said elements from said subsequent interval which are anomalous because they differ by a criteria from comparable elements of said current interval of elements wherein compressed measurement data correlative to machine status is continuously captured.

36. The device of claim 35 further including means for replacing only those elements in said current interval which correspond to said anomalous elements from said subsequent interval to form a subsequent current interval of elements for future comparison by said comparing means.

37. A method for compressing measurement data correlative to machine status, the steps including:

sensing machine data correlative to machine status;

sampling said data;

transforming said sampled data into spectral elements;

comparing said spectral elements to a user definable criteria for retention;

compressing said sampled data by storing in a memory means those spectral elements which have passed said user definable criteria for retention wherein said stored spectral elements are correlative to machine status;

sensing a mechanical phase reference mark, and

demarcating said sensed machine data into periods having a length correlative to said sensed mechanical phase reference mark.

38. A signal processing method for processing machinery signals correlative to machine status, the steps including:

sensing signals correlative to machine status;

converting said signals into digital values;

storing a series of said digital values into packets;

comparing a subsequent packet of said digital values with at least one said stored packet of said digital values;

storing said digital values included in said subsequent packet which are anomalous because they differ by a criteria from digital values of said at least one stored packet which has been compared with said subsequent packet;

transmitting a signal correlative to said subsequent packet of digital values including flagging anomalous data in said subsequent packet of digital values.

39. A signal processing method for processing machinery signals correlative to machine status, the steps including:

sensing data correlative to machine status;

sampling the data;

transforming the sampled data into periods of spectral elements;

storing a first period of the spectral elements in a memory defining a previous period of stored spectral elements;

comparing a subsequent period of spectral elements to the previous period for determining anomalous behavior between corresponding spectral elements;

using the anomalous spectral elements from the subsequent period to replace the corresponding spectral elements in the previous period for defining a new previous period of spectral elements.

40. The method of claim 39 further including the step of iteratively updating the new previous period of spectral elements by replacing spectral elements in the new previous period with any anomalous spectral elements determined each time further subsequent periods of spectral elements are compared with the new previous period of spectral elements.

41. The method of claim 39 further including the step of transmitting the first period of the spectral elements and the anomalous spectral elements of the subsequent period to a remote memory for storing the first period of the spectral elements and the anomalous spectral elements at identified positions of each respective period.

42. The method of claim 41 further including the step of iteratively updating the new previous period of spectral elements by replacing spectral elements in the new previous period with any anomalous spectral elements determined each time further subsequent periods of spectral elements are compared with the new previous period of spectral elements.

43. The method of claim 42 further including the step of transmitting to the remote memory anomalous spectral elements determined each time further subsequent periods of spectral elements are compared with the new previous period of spectral elements and storing the determined anomalous spectral elements at identified positions of each respective period.

44. The method of claim 43 further including the step of reconstituting any period of spectral elements by starting at a period of spectral elements that is to be reconstituted and sequencing backwards through the periods of spectral elements stored in the remote memory and using the spectral elements stored at identified positions in the periods which are different in position than the identified positions for the period being reconstituted.

45. A method for compressing data correlative to machine status, the steps including:

sensing data correlative to machine status;

sampling said data;

transforming said sampled data into a plurality of sets of spectral elements;

comparing a first set in the plurality of sets of spectral elements to a user definable criteria for retention;

storing in a memory those spectral elements which have passed the criteria for retention;

comparing a subsequent set of spectral elements in the plurality of sets of spectral elements to the criteria, and comparing those spectral elements which have passed the criteria for retention to the first set of spectral elements stored in memory

compressing data by storing in the memory those spectral elements included in the subsequent set which are anomalous because they differ by a user definable amount from comparable elements of the first set of spectral elements stored in memory.

46. A signal processing method for processing machinery signals correlative to machine status, the steps including:

sensing data correlative to machine status;

sampling the data;

transforming the sampled data into periods of spectral elements;

storing a first period of the spectral elements in a memory defining a previous period of stored spectral elements;

comparing a subsequent period of spectral elements to the previous period of stored spectral elements for determining anomalous data between corresponding spectral elements;

transmitting a signal correlative to the subsequent period of spectral elements including flagging the anomalous data in the subsequent period of spectral elements.